Data visualisation

• visualising our data helps us visualise the relationship between variables in order to tell a story
• we typically visualise variables that we have a particular hypothesis for: predictor and measure variable(s)

Visualising distributions

• histograms, density plots, and bar plots for count values all visualise the distribution of observations
  • they tell us about how many times we observed certain values of a variable
  • we usually do this to get a feel for what our data look like
    • what is the range of our data, the mode, the overall distribution of values?

Visualising relationships

• to visualise relationships between variables, we need to have at least two variables mapped to a plot’s aesthetics
• we’ve already done this by mapping colour or fill to a categorical variable, while mapping: + a continuous variable to the x-axis for histograms/density plots, or
  • a categorical variable to the y-axis for a barplot

Grouped continuous variable

• our histograms, density plots, and bar plots all shows the distribution of values of a continuous variable by different levels of a categorical variable

Stacked

• recall that by default these categories are “stacked” on top of each other

Abbildung 1: Visualising relationships in distributions

Dodged

• but that we can have them side-by-side by using the ‘dodge’ function
  • I find this is more useful for bar plots

Abbildung 2: Visualising relationships in distributions

Two continuous variables

• we often want to see the effect one continuous variable has on another
• for example, in our english dataset we have the variables WrittenFrequency and RTlexdec
  • what type of relationship do you think these two variables will have?
  • i.e., do you think that words with lower WrittenFrequency will tend of have longer or shorter reaction times in a lexical decision task?
  • how might we visualise such a relationship?

Adding more variables

• recall that we can use other aesthetics like fill or colour
  • for geom_point(), it’s also helpful to use shape

df_english |> 
  ggplot() +
  aes(x = WrittenFrequency, y = RTlexdec,
      colour = AgeSubject,
      shape = AgeSubject) +
  geom_point()

• there’s a lot of overlap in the middle of the plot
  • how can we change the opacity of the points?

Facet grids

• in general, if you want to plot more than three variables it’s a good idea to split categorical variables into facets
  • facets are subplots that displays subsets of the data
• we can use facet_wrap(), which takes a formula as its argument
  • this formula includes ~ and the name of a cateogircal variable, e.g., ~CV

Wrangled data

• We can also wrangle our data before feeding it into ggplot()
  • This is useful when we don’t want to make permanent changes to the data, but we just want to plot for example a subset of the data
• maybe we only want to look at words that begin with a vowel, how might we do this with a dplyr verb?

Quarto code chunk settings

• long code chunks can lead to very messy output documents
• usually only the plot is important to the reader, not the code that produced it
• we can control the presentation and evaluation of code chunks through code chunk options
  • these begin with #|
  • and are placed directly below ```{r}```
• important code chunk options:

Using code chunks

• why do we not see the result of this plot?

```{r}
#| eval: false
df_english |> 
  ggplot() +
  aes(x = RTlexdec, y = RTnaming,
      colour = AgeSubject,
      shape = AgeSubject) +
  geom_point()
```

Saving plots

• we often want to use our plots in a document that is not created in RStudio

  • for example in a thesis or a paper written in LaTeX

• to do this we need to load in our figures as an accepted file type, such as jpeg or png

• we can do this with the ggsave() function

• Can you guess what types of arguments ggsave() needs in order to save our plots? Some are required, some are optional.

ggsave()

At minimum ggsave() takes as its arguments:

1. the name of the plot in your Environment that you wish to save
2. the file name you wish to save your plot under
   • it’s a good idea to create a folder where you will save your plots, and to include the filepath in the name

ggsave() optional argumets

• some optional arguments are:
  • width =: how wide you want your plot to be in cm, mm, inches, or pixels
  • height =:
  • dpi =: desired resolution (numerical, or a set of strings: “retina” = 320, “print” = 300, or “screen” = 72)
  • bg =: background colour, e.g., “black”